List your company

Hallucination Evaluation on CRPE relation

79.1AccuracyInternVL2.5-26B
56.63662.46868.374.132Dec 6, 2024Feb 14, 2025Apr 26, 2025Jul 6, 2025Sep 14, 2025Nov 24, 2025Feb 3, 2026

Evaluation results

MethodLinks
InternVL2.5-26BModel Name=InternVL2.5-26B2024.1279.1
InternVL2.5-78BModel Name=InternVL2.5-78B2024.1278.8
InternVL2.5-8BModel Name=InternVL2.5-8B2024.1278.4
InternVL2.5-38BModel Name=InternVL2.5-38B2024.1278.3
Qwen3-VL + DISCO (10K)Backbone=Qwen3-VL, Alignment Strategy=DISCO (10K)2026.0277.92
Qwen3-VL + Textual (10K)Backbone=Qwen3-VL, Alignment Strategy=Textual (10K)2026.0277.85
Qwen3-VLBackbone=Qwen3-VL2026.0277.68
InternVL2-40BModel Name=InternVL2-40B2024.1277.6
InternVL2-Llama3-76BModel Name=InternVL2-Llama3-76B2024.1277.6
GPT-4o-20240513Model Name=GPT-4o-202405132024.1276.6
MiniCPM-v2.6 + HIPPOBackbone=MiniCPM-v2.6, Alignment Strategy=HIPPO2026.0276.37
MiniCPM-v2.6Backbone=MiniCPM-v2.62026.0276.32
InternVL2-8BModel Name=InternVL2-8B2024.1275.8
InternVL2-26BModel Name=InternVL2-26B2024.1275.6
InternVL2.5-4BModel Name=InternVL2.5-4B2024.1275.5
InternVL-Chat-V1.5Model Name=InternVL-Chat-V1.52024.1275.4
MiniCPM-V2.6Model Name=MiniCPM-V2.62024.1275.2
Qwen2-VL-7BModel Name=Qwen2-VL-7B2024.1274.4
InternVL2-4BModel Name=InternVL2-4B2024.1271.1
InternVL2.5-2BModel Name=InternVL2.5-2B2024.1270.2
InternVL2-2BModel Name=InternVL2-2B2024.1266.3
InternVL2.5-1BModel Name=InternVL2.5-1B2024.1260.9
InternVL2-1BModel Name=InternVL2-1B2024.1257.5